Related Experiment Video
Updated: Jan 9, 2026
Predictive Immune Modeling of Solid Tumors
Published on: February 25, 2020
Integrative prognostic model incorporating high mobility group box 1 subcellular localization and tumor-infiltrating
Yao Liu1,2,3, Miyako Shimasaki4, Motona Kumagai5,6
1Department of Pathology and Laboratory Medicine, Kanazawa Medical University, Ishikawa, Uchinada-machi, 920-0293, Japan.
Background:
The prognostic impact of high mobility group box 1 (HMGB1) in lung adenocarcinoma may depend on its subcellular localization, while the density of tumor-infiltrating lymphocytes (TILs) reflects the host anti-tumor immune response. However, the combined prognostic value of these two factors in early-stage lung adenocarcinoma remains unclear.
Methods:
This retrospective study included 112 patients with pathological stage I-II lung adenocarcinoma who underwent complete surgical resection at our institution between 2007 and 2017. None received neoadjuvant chemotherapy or radiotherapy. Immunohistochemistry was performed on formalin-fixed, paraffin-embedded tumor specimens to evaluate HMGB1 subcellular localization and stromal TILs infiltration. The latter was semi-quantitatively assessed according to the International TILs Working Group recommendations and dichotomized using the median value as the cutoff. Clinicopathological variables, including differentiation, pleural invasion, lymphovascular invasion, and pathological stage, were collected and correlated with HMGB1 localization and TIL status. Survival outcomes were analyzed using Kaplan-Meier and Cox proportional hazards models. Multivariable analyses were adjusted according to the events-per-variable (EPV) rules, and model diagnostics included proportional hazards testing and multicollinearity assessment. Interobserver agreement for HMGB1 localization was evaluated using Fleiss'κ statistics.
Results:
Cytoplasmic HMGB1 expression and low TIL infiltration were significantly associated with adverse clinicopathological features, including poorer differentiation and higher rates of lymphovascular invasion. Both cytoplasmic HMGB1 and low TIL levels independently predicted a shorter DFS and OS. Patients with the combined phenotype of cytoplasmic HMGB1 and low TIL levels had the worst prognosis, with hazard ratios exceeding those of either factor alone. The integrative model based on HMGB1 localization and TIL status enhanced the prognostic discrimination beyond conventional clinicopathological parameters.
Conclusions:
HMGB1 subcellular localization and TIL infiltration are independent prognostic biomarkers of early-stage lung adenocarcinoma. An integrative model combining these parameters provides enhanced risk stratification and may inform individualized postoperative management strategies.
